Research On The Fault Diagnosis System Of Armed Vehicles

The Engine is an extremely important part of the Armed vehicles,research on its fault diagnosis technology has gotten more and more attention.As acoustical diagnosis is a non-contact method,it also draws more attention because of its easy operation.In this paper,the principle of several common failure and their characteristics of acoustic signal about the engine of a particular armored vehicles are concretely analyzed by frequency and time domain analysis,such as firing advance angle in advance/lag,the gap of air-in oversize,the gap of air-out oversize and one tank not working and so on.According to the non-stationary characteristic of acoustic signals, the Wavelet Packet Decomposition which is applicable to non-stationary signal analysis is used to analyse the fault signal,the Decomposition of these signals based on the wavelet Decomposition shows that a minutia of signals can be magnified.The difference between them can be identified by comparing the frequency with that of fault signals.De-noising function of the non-linear Wavelet Decomposition theory is improved,using the layered threshold de-noising method to process the fault signals, and its great effectiveness of de-noising is verified.By researching wavelet transform-threshold,the layered threshold de-noising method which based on nonlinearity-wavelet transform is improved,the method is perfect compared to other nonlinearity-wavelet transform threshold methods.It is testified by experiment that the method can remarkably improve the precision of filtering,at the same time,it can effectively keep the primary minutia of the signal.Combining wavelet Packet Decomposition with energy spectrum analysis method,it is proposed that the interval wavelet packet which is based on partial-band wavelet packet energy is used to extract the signal feature.The basic principles are based on energy distribution of the characteristics of the signals in a series of time-frequence and resolving space which effectively portray the inherent,and prominent band of the key feature can be extracted.In order to achieve intelligent fault diagnosis of a armed vehicles,using Wavelet neural network to identify,and the reformative Wavelet Decomposition and BP neural networks to diagnose the fault from the acoustical signal of engine.Then the speed join in the feature vector,considering it affecting the feature value of the fault diagnosis,that is,the feature vector is composed of the feature from wavelet packet and the speed.Whereafter sending the feature vector from wavelet packet to BP neural networks,it is tested that the method can effectively achieve the intelligent fault diagnosis.